Using the illustration in Test for the Same Distribution Across Groups, it would likely be best to create vectors instead of matrices if the group sizes are not equal, for example:
x = [sales_A; sales_B];
group = [ones(size(sales_A)); 2*ones(size(sales_B))];
Since ‘Sales’ appears to be a cell array of column vectors, I am assuming that ‘x’ and ‘group’ will both be column vectors as well. (Change this and the code if my guess is incorrect.)
To use a matrix argument, both columns need to have the same row size, and since they are apparently not the same size, a matrix will not work here. They have to be vectors, creating a new vector.
I cannot test this without ‘Sales’, so I am posting it as UNTESTED CODE.
