Perform student t-test on 3D matrix (on each grid point)

Hi I have a 3D matrix (x,y,z) and I need to perform a t-test at each grid point (x,y) to test if the mean of the values in the z direction is significantly higher than zero.

My attempt:

A=data(306,6,63); %data
for x=1:306
  for y=1:6
    h(x,y)=ttest(A(x,y,:),'Alpha',0.01,'Tail','right'):
  end
end

I wasn't sure how to handle the z direction in the loop. I've played around this this a lot getting various errors, right now it says "The data in a paired t-test must be the same size". I want to do a one-sample test on each grid point to get a 2D output indicating which grid points are significantly different.

Answer 1

[the cyclist]

It's just a minor syntax error. Try

h(x,y)=ttest(A(x,y,:),[],'Alpha',0.01,'Tail','right');

Notice that I put in an empty array for the second argument.

(This was not particularly clear, in my opinion, from the documentation.)

[Jaclyn]

Thanks for your reply, I put in the square brackets and now I'm getting the error: Too many input arguments.

[the cyclist]

Since I did not have your data, this is the code I ran:

A = randn(306,6,63);

for x=1:306
  for y=1:6
    h(x,y)=ttest(A(x,y,:),[],'Alpha',0.01,'Tail','right');
  end
end

Does that work for you?

If not, it could be a versioning issue. I have R2013a on Mac OS X 10.8.4.

[Jaclyn]

Ah yes that's probably the issue! Never thought of that, I have 2011. I just looked up the old documentation and I got this to work:

for x=1:306
  for y=1:6
    h(x,y)=ttest(A(x,y,:),0,0.01,'right');
  end
end

Thanks for your help!