fast_t2

 fast_t2 - Computes a 2D matrix of independent sample t-tests relatively 
           quickly.

 Usage:
 [p_values, t_scores, mn_dif, stder]=fast_t2(dataA,dataB,tail,verblevel);

 Inputs:
  dataA   - 3D matrix of data (Channel x Time x Participant) for Group A
  dataB   - 3D matrix of data (Channel x Time x Participant) for Group B

 Optional Inputs:
  tail        - [1 | 0 | -1] If tail=1, the alternative hypothesis is that
                the mean of Group A is greater than that of Group B.  If tail=0,
                the alternative hypothesis is that the mean of the groups are different
                than 0 (two tailed test).  If tail=-1, the alternative hypothesis
                is that the mean of Group A is less than that of Group B.
                {default: 0}
  verblevel   - An integer specifiying the amount of information you want
                this function to provide about what it is doing during runtime.
                 Options are:
                    0 - quiet, only show errors, warnings, and EEGLAB reports
                    1 - stuff anyone should probably know
                    2 - stuff you should know the first time you start working
                        with a data set {default value}
                    3 - stuff that might help you debug (show all
                        reports)

 Outputs:
  p_values    - p-value of difference between groups at each time point 
                and electrode (no correction for multiple comparisons)
  t_scores    - t-score of difference between groups at each time point 
                and electrode
  mn_dif      - mean voltage difference between groups at each time point 
                and electrode (Group A-Group B)
  stder       - standard error of the mean voltage difference at each time
                point and electrode

 Author:
 David Groppe
 May, 2010
 Kutaslab, San Diego

0001 % fast_t2 - Computes a 2D matrix of independent sample t-tests relatively
0002 %           quickly.
0003 %
0004 % Usage:
0005 % [p_values, t_scores, mn_dif, stder]=fast_t2(dataA,dataB,tail,verblevel);
0006 %
0007 % Inputs:
0008 %  dataA   - 3D matrix of data (Channel x Time x Participant) for Group A
0009 %  dataB   - 3D matrix of data (Channel x Time x Participant) for Group B
0010 %
0011 % Optional Inputs:
0012 %  tail        - [1 | 0 | -1] If tail=1, the alternative hypothesis is that
0013 %                the mean of Group A is greater than that of Group B.  If tail=0,
0014 %                the alternative hypothesis is that the mean of the groups are different
0015 %                than 0 (two tailed test).  If tail=-1, the alternative hypothesis
0016 %                is that the mean of Group A is less than that of Group B.
0017 %                {default: 0}
0018 %  verblevel   - An integer specifiying the amount of information you want
0019 %                this function to provide about what it is doing during runtime.
0020 %                 Options are:
0021 %                    0 - quiet, only show errors, warnings, and EEGLAB reports
0022 %                    1 - stuff anyone should probably know
0023 %                    2 - stuff you should know the first time you start working
0024 %                        with a data set {default value}
0025 %                    3 - stuff that might help you debug (show all
0026 %                        reports)
0027 %
0028 % Outputs:
0029 %  p_values    - p-value of difference between groups at each time point
0030 %                and electrode (no correction for multiple comparisons)
0031 %  t_scores    - t-score of difference between groups at each time point
0032 %                and electrode
0033 %  mn_dif      - mean voltage difference between groups at each time point
0034 %                and electrode (Group A-Group B)
0035 %  stder       - standard error of the mean voltage difference at each time
0036 %                point and electrode
0037 %
0038 % Author:
0039 % David Groppe
0040 % May, 2010
0041 % Kutaslab, San Diego
0042 
0043 %%%%%%%%%%%%%%%% REVISION LOG %%%%%%%%%%%%%%%%%
0044 %
0045 
0046 
0047 function [p_values, t_scores, mn_dif, stder]=fast_t2(dataA,dataB,tail,verblevel)
0048 
0049 if nargin<2,
0050     error('You need to provide data.');
0051 end
0052 
0053 if nargin<3,
0054     tail=0; %default two-tailed test
0055 elseif (tail~=0) && (tail~=1) && (tail~=-1),
0056     error('Argument ''tail'' needs to be 0,1, or -1.');
0057 end
0058 
0059 if nargin<4,
0060    verblevel=2; 
0061 end
0062 
0063 [n_chanA, n_ptsA, n_subsA]=size(dataA);
0064 [n_chanB, n_ptsB, n_subsB]=size(dataB);
0065 if n_chanA~=n_chanB,
0066     error('dataA and dataB have a different number of channels. They need to be the same.')
0067 end
0068 if n_ptsA~=n_ptsB,
0069     error('dataA and dataB have a different number of time points. They need to be the same.')
0070 end
0071 
0072 df=n_subsA+n_subsB-2;
0073 if verblevel~=0,
0074     fprintf('fast_t2: Number of channels: %d\n',n_chanA);
0075     fprintf('fast_t2: Number of time points: %d\n',n_ptsA);
0076     fprintf('fast_t2: Total # of comparisons: %d\n',n_ptsA*n_chanA);
0077     fprintf('fast_t2: Number of participants in Group A: %d\n',n_subsA);
0078     fprintf('fast_t2: Number of participants in Group B: %d\n',n_subsB);
0079     fprintf('t-score degrees of freedom: %d\n',df);
0080 end
0081 
0082 smA=sum(dataA,3);
0083 mnA=smA/n_subsA;
0084 ssA=sum(dataA.^2,3)-(smA.^2)/n_subsA;
0085 
0086 smB=sum(dataB,3);
0087 mnB=smB/n_subsB;
0088 ssB=sum(dataB.^2,3)-(smB.^2)/n_subsB;
0089 
0090 mult_fact=(n_subsA+n_subsB)/(n_subsA*n_subsB);
0091 pooled_var=(ssA+ssB)/df;
0092 stder=sqrt(pooled_var*mult_fact);
0093 
0094 mn_dif=mnA-mnB;
0095 t_scores=mn_dif./stder;
0096 
0097 if tail<0,
0098     %lower tailed test
0099     p_values=tcdf(t_scores,df);
0100 elseif tail>0,
0101     %upper tailed test
0102     p_values=1-tcdf(t_scores,df);
0103 else
0104     %two tailed test
0105     p_values=tcdf(t_scores,df);
0106     ids=find(p_values>.5); %t-scores above zero
0107     p_values(ids)=1-p_values(ids); 
0108     p_values=p_values*2; %double for two tailed test
0109 end
0110   
0111