Matrix Input to bootci

According to the documentation for bootci , the input data can be a matrix, and the function will compute a confidence interval over the rows. So for example, if my input data is:

>> X = randn(10,200,4);

the expected behavior of the function is that it should output a matrix that is [2 200 4] (lower/upper bounds for each of the 200x4 data).

However, when I call the function:

>> ci = bootci(1000,{@nanmean,X},'alpha',0.05,'type','norm');

I get an error:

Error using bsxfun
Non-singleton dimensions of the two input arrays must match each other.

Error in bootci>bootnorm (line 190)
bias = mean(bsxfun(@minus,bstat,stat),1);

Error in bootci (line 161)
      [ci,bootstat] = bootnorm(obsstat,nboot,fun,alpha,weights,bootstrpOptions,data{:});

Possibly related, the output of the function with a 2D matrix is also unexpected given the documentation:

>> X = randn(10,200);
>> ci = bootci(1000,{@nanmean,X},'alpha',0.05,'type','norm');

>> size(ci)

2 10

If the function were operating on the rows (as it says in the help file), it should give a [2 200] output.

Has anyone noticed this behavior and/or gotten it to work properly in both cases?

Thanks,

Matt

Answer 1

[Matt]

EDIT: Problem solved. I had an external version of nanmean in my path. The built-in function works as expected.