pr_hoch

 Step up correction for multiple comparisons.
 FORMAT H = pr_hoch(p,alfa,No,sortf)

 Input:
 p       - pvalues
 alfa    - FWE
 No      - number of Null Hypotheses
 sortf   - is 1 if p is already sorted ascending

 Output:
 H       - vector of rejections 
           1 hypothesis rejected
           0 otherwise
  
 References:
 Hochberg Y (1988) "A sharper Bonferroni procedure for 
 multiple tests of significance," Biometrika 75:800-803 

 Hochberg Y, Benjamini Y (1990) "More powerful procedures for 
 multiple significance testing," Statist.  Med. 9:811-818

 Federico E. Turkheimer 15/6/2000

 $Id: pr_hoch.m,v 1.1 2005/06/01 09:24:23 matthewbrett Exp $

0001 function H = pr_hoch(p,alfa,No,sortf)
0002 % Step up correction for multiple comparisons.
0003 % FORMAT H = pr_hoch(p,alfa,No,sortf)
0004 %
0005 % Input:
0006 % p       - pvalues
0007 % alfa    - FWE
0008 % No      - number of Null Hypotheses
0009 % sortf   - is 1 if p is already sorted ascending
0010 %
0011 % Output:
0012 % H       - vector of rejections
0013 %           1 hypothesis rejected
0014 %           0 otherwise
0015 %
0016 % References:
0017 % Hochberg Y (1988) "A sharper Bonferroni procedure for
0018 % multiple tests of significance," Biometrika 75:800-803
0019 %
0020 % Hochberg Y, Benjamini Y (1990) "More powerful procedures for
0021 % multiple significance testing," Statist.  Med. 9:811-818
0022 %
0023 % Federico E. Turkheimer 15/6/2000
0024 %
0025 % $Id: pr_hoch.m,v 1.1 2005/06/01 09:24:23 matthewbrett Exp $
0026 
0027 if nargin < 1
0028   error('Need p values');
0029 end
0030 if nargin < 2
0031   alfa = 0.05;
0032 end
0033 n    = length(p);
0034 if nargin < 3
0035   No = n;
0036 elseif No > n
0037   No = n;
0038 end
0039 if nargin < 4
0040   sortf = 0;
0041 end
0042 if ~sortf
0043   q = sort(p(:));
0044 else
0045   q = p(:);
0046 end
0047 
0048 H1 = q < alfa./(min(No, (n:-1:1)'+1));
0049 th = max([q(H1); 0]);
0050 H = p<= th;
0051 
0052 return