Say for example I calculated two sets of cross-correlation matrices, one that compared time series A1 and A2 for 10 test subjects and another that compared time series B1 and B2 for the same 10 test subjects. If I took the average of each of these cross-correlation sets, I would have the average matrix A for all cross-correlations between A1 and A2, and the average matrix B for all cross-correlations between B1 and B2:

Lag = [   -3 lag;      -2 lag;      -1 lag;       0 lag;     +1 lag;      +2 lag;      +3 lag]
A = [0.003100173; 0.042535000; 0.140213224; 0.256917007; 0.17103877; 0.095175853; 0.062051573]
B = [0.030890086; 0.023768139; 0.025463862; 0.086639792; 0.03560601; 0.012300708; 0.019001531]

Is there a way to then compare the similarity of these two average cross-correlation matrices? I essentially want to say something about whether the cross-correlation is statistically stronger between time series A1 and A2 versus time series B1 and B2 at a time lag of 0 (which would be 0.256917007 for A and 0.086639792 for B).

I thought about just doing a dependent-samples t-test to check for differences between A and B since I have variance and standard error around these averages, but that feels almost too simple? Should I be looking into chi-square to see whether the number of people who had maximum cross-correlation at 0 were different between A and B? Or are there more sophisticated analyses that would be better here? I would love any help or advice or resources that would be good for understanding what I can do here!



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.