Why degrees of freedom is not correctly represented in repeated measure ANOVA (ranova) table?

Hello,

I am trying to do a repeated measure ANOVA using the ranova command in 2014 Matlab. I am trying to compare activity of a number of cells (lets say 60, as random variables) across 2 different conditions through 21 epochs of time. Here is how I define my table:

Epoch=[(1:21)';(1:21)'];

Task(1:21,1)=['A'];

Task(22:42,1)=['B'];

Task_Epoch=table(Task,Epoch);

cellnum=[1:60]';

RespTable = table(cellnum,Pref_act_Epoch(1,:)',Pref_act_Epoch(2,:)',Pref_act_Epoch(3,:)',Pref_act_Epoch(4,:)',Pref_act_Epoch(5,:)',Pref_act_Epoch(6,:)',Pref_act_Epoch(7,:)',Pref_act_Epoch(8,:)',Pref_act_Epoch(9,:)',Pref_act_Epoch(10,:)',Pref_act_Epoch(11,:)',Pref_act_Epoch(12,:)',Pref_act_Epoch(13,:)',Pref_act_Epoch(14,:)',Pref_act_Epoch(15,:)',Pref_act_Epoch(16,:)',Pref_act_Epoch(17,:)',Pref_act_Epoch(18,:)',Pref_act_Epoch(19,:)',Pref_act_Epoch(20,:)',Pref_act_Epoch(21,:)',Pref_actPID_Epoch(1,:)',Pref_actPID_Epoch(2,:)',Pref_actPID_Epoch(3,:)',Pref_actPID_Epoch(4,:)',Pref_actPID_Epoch(5,:)',Pref_actPID_Epoch(6,:)',Pref_actPID_Epoch(7,:)',Pref_actPID_Epoch(8,:)',Pref_actPID_Epoch(9,:)',Pref_actPID_Epoch(10,:)',Pref_actPID_Epoch(11,:)',Pref_actPID_Epoch(12,:)',Pref_actPID_Epoch(13,:)',Pref_actPID_Epoch(14,:)',Pref_actPID_Epoch(15,:)',Pref_actPID_Epoch(16,:)',Pref_actPID_Epoch(17,:)',Pref_actPID_Epoch(18,:)',Pref_actPID_Epoch(19,:)',Pref_actPID_Epoch(20,:)',Pref_actPID_Epoch(21,:)');

R1 = fitrm(RespTable,'Var2-Var43~cellnum','WithinDesign',Task_Epoch);

ranovatbl =ranova(R1,'WithinModel','Task+Epoch+Task*Epoch');

The RespTable looks exactly as it should and the ranova table looks reasonable too, however, the df is always 1 whether it is looking at the task, or epoch or even their interception. Anybody knows why and what I am doing wrong?

P.S. Is there any more civilized way of filling the Resptable rather than naming all the 42 epochs one by one?

Regards, Mohsen

Answer 1

[Sophia Snipes]

Hello Mohsen,

I know this is way too late for you, but I found your question because I had the same one, and was eventually able to find the answer, at least to my problem.

For the within-design table, MATLAB treats the values differently if it's numerical or categorical. so in your case, before inserting "Epoch" into the Task_Epoch table, you have to make it categorical first:

Epoch = categorical(Epoch);

[Mohsen]

Thanks for your great explanation.