In the record of “From correlation coefficients to variance components”, slide 15.

it was mentioned that in the model, Y = X b + e ; b follows a normal distribution of N(0, I . sG2 / m ).

I want to know what is I, S and G2?

Good question, the notation I used could be clearer.

I is the identity matrix (a NxN matrix with 1 on the diagonal and 0 otherwise), sG2 is a number and corresponds to the variance of the random effect (square of the standard deviation). In several papers it is called sigma_G squared, which is more self explanatory but I got a bit lazy typing the equations.

1 Like