I have two correlation (R Pearson) matrices with same dimensions (i.e. equal number of columns and rows). Does anyone know any tests/statistics that I can apply in order to compare them?
My Dear.. You can use p-value of each correlation which is small is high different for example:
if r11(1) for first matrix with p-value(sig.=0.003), and r11(2) from second matrix with (sig.=0.024) each them are signficant, but r11(1) is very strong compared with r11(2) becuase p-value_1
My Dear.. You can use p-value of each correlation which is small is high different for example:
if r11(1) for first matrix with p-value(sig.=0.003), and r11(2) from second matrix with (sig.=0.024) each them are signficant, but r11(1) is very strong compared with r11(2) becuase p-value_1
The standard test is given in Morrison, D. F., Multivariate Statistical Methods, 4th ed. Duxbury/Thomson. in Section 5.4 (Testing the Equality of Several Covariance Matrices,), p. 247 ff
You might even consider applying a permutation test approach: by randomly assigning your observations to two groups of the same size as the original groups, each time calculating the resulting difference in covariation values. When repeating this a sufficient number of times (say 10.000 repeats), the difference between the original covariation matrices is significant if less than 5 per cent of the resulting differences are larger than the one you obtained.