function [s, cfg] = ft_statfun_diff_itc(cfg, dat, design)

% FT_STATFUN_DIFF_ITC computes the difference in the inter-trial coherence between

% two conditions. The input data for this test should consist of complex-values

% spectral estimates, e.g. computed using FT_FREQANALYSIS with cfg.method='mtmfft',

% 'wavelet' or 'mtmconvcol'.

%

% The ITC is a measure of phase consistency over trials. By randomlly shuffling the

% trials between the two consitions and repeatedly computing the ITC difference, you

% can test the significance of the two conditions having a different ITC.

%

% A difference in the number of trials poer condition will affect the ITC, however

% since the number of trials remains the same for each random permutation, this bias

% is reflected in the randomization distribution.

%

% Use this function by calling the high-level statistic functions as

% [stat] = ft_freqstatistics(cfg, freq1, freq2, ...)

% with the following configuration option:

% cfg.statistic = 'ft_statfun_diff_itc'

%

% For this specific statistic there is no known parametric distribution, hence the

% probability and critical value cannot be computed analytically. This specific

% statistic can therefore only be used with cfg.method='montecarlo'. If you want to

% do this in combination with cfg.correctm='cluster', you also need to specify

% cfg.clusterthreshold='nonparametric_common' or 'nonparametric_individual'.

%

% You can specify the following configuration options:

% cfg.complex = string, 'diffabs' (default) to compute the difference of the absolute ITC values,

% or 'absdiff' to compute the absolute value of the difference in the complex ITC values.

%

% The experimental design is specified as:

% cfg.ivar = independent variable, row number of the design that contains the labels of the conditions to be compared (default=1)

%

% The labels for the independent variable should be specified as the number 1 and 2.

%

% See also FT_FREQSTATISTICS and FT_STATISTICS_MONTECARLO

% Copyright (C) 2008-2014, Robert Oostenveld

%

% This file is part of FieldTrip, see http://www.fieldtriptoolbox.org

% for the documentation and details.

%

% FieldTrip is free software: you can redistribute it and/or modify

% it under the terms of the GNU General Public License as published by

% the Free Software Foundation, either version 3 of the License, or

% (at your option) any later version.

%

% FieldTrip is distributed in the hope that it will be useful,

% but WITHOUT ANY WARRANTY; without even the implied warranty of

% MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the

% GNU General Public License for more details.

%

% You should have received a copy of the GNU General Public License

% along with FieldTrip. If not, see <http://www.gnu.org/licenses/>.

%

% $Id$

% set the defaults

cfg.complex = ft_getopt(cfg, 'complex', 'diffabs');

cfg.ivar = ft_getopt(cfg, 'ivar', 1);

sel1 = find(design(cfg.ivar,:)==1);

sel2 = find(design(cfg.ivar,:)==2);

df1 = length(sel1);

df2 = length(sel2);

if (df1+df2)<size(design, 2)

ft_error('inappropriate design, should contain 1''s and 2''s');

end

if isreal(dat)

ft_error('the data should be complex, i.e. computed with FT_FREQANALYSIS and cfg.output="fourier"');

end

% normalise the complex data in each trial

dat = dat./abs(dat);

switch cfg.complex

case 'diffabs'

% first compute the absolute, then take the difference

% this is not sensitive to phase differences

itc1 = abs(mean(dat(:,sel1), 2)); % ITC is the length of the average complex number

itc2 = abs(mean(dat(:,sel2), 2)); % ITC is the length of the average complex number

s.stat = itc1 - itc2;

case 'absdiff'

% first compute the difference, then take the absolute

% this is sensitive to phase differences

itc1 = mean(dat(:,sel1), 2); % ITC is here the average complex number

itc2 = mean(dat(:,sel2), 2); % ITC is here the average complex number

s.stat = abs(itc1 - itc2);

otherwise

ft_error('incorrect specification of cfg.complex');

end