anovanTable

Calls ANOVAN, but takes an N-dimensional array as the data.

[p, table, stats] = anovanTable(T, varargin)

Syntax exactly as ANOVAN, except X and Y of the linear model are replaced by an N-dimensional array.

Each dimension becomes one factor of the ANOVA design, so if 10 individuals were tested on 2 conditions at 3 timepoints, one would create a 10 x 2 x 3 array.

example:

T = randn([10 2 3]);
p = anovanTable(T, ...
'varnames', {'Individual', 'Condition','Time'}, ...
);

this is identical to converting your table to a vector Y = T(:), and creating a design matrix X where each row contains the table subscripts of the corresponding item in vector Y.

Repeated measures can be achieved by adding an extra dimension to the array, and using the parameters ('collapse', DIM). If there are different numbers of measurements in each condition, then the extra dimensions can be padded with NaN.

To see the actual arguments that get passed to ANOVAN, you can get the 5th and 6th return values, which correspond to Y and X respectively.