T-Test2 on a singleton dimension

I am conducting a two-sample t-test on some self-organizing maps data. I am trying to obtain p-values at every grid cell between each node:

somVariable = sMap.codebook % 30 x7326 [double] ; node x [latxlon]

[sNum,sGrid] = size(somVariable); 

pTest = NaN(sGrid,sNum);
for i = 1:sNum
    [~,p] = ttest2(somVariable(1,:),somVariable(i,:),'VarType','equal');
    pTest(:,i) = p';
    clear p
end

but encounter an issue in which matlab jumps to the non-singleton dimension (7326) and I obtain one p-value. Is there a way to force matlab to recognize a singleton dimension as the starting dimension to test at each grid cell?

Thanks!

[Scott MacKenzie]

I see some issues in your code and question. For example, ttest2 only returns one p-value. So, p and p' are the same.

[the cyclist]

@Manny Hernandez, I'm confused by which elements of your data you are trying to apply the t-test to. Let's take a small example. Suppose somVariable were:

somVariable = [ 2  3  5  7;
               11 13 17 19;
               23 29 31 37];

Exactly which t-tests are you trying to do? The way your code is written, you calculate three p-values:

• Row 1 vs. Row 1 (p = 1)
• Row 1 vs. Row 2 (p = 0.0024)
• Row 1 vs. Row 3 (p = 0.0002)

and then those three p-values are replicated four times (because you have defined pTest to have as many rows as somVariable has columns).

somVariable = [ 2  3  5  7;
               11 13 17 19;
               23 29 31 37];

[sNum,sGrid] = size(somVariable); 

pTest = NaN(sGrid,sNum);
for i = 1:sNum
    [~,p] = ttest2(somVariable(1,:)',somVariable(i,:)','VarType','equal');
    pTest(:,i) = p';
    clear p
end

disp(pTest)

    1.0000    0.0024    0.0002
    1.0000    0.0024    0.0002
    1.0000    0.0024    0.0002
    1.0000    0.0024    0.0002