I am trying to compute the variance of the multiplicative multinomial distribution. I know that it can be computed using the MOMENT DEFINITION but I don't understand which kind of technique this is. I have never heard about it.
In particular the multiplicative multinomial has a density function like $$P(\textbf{Y} = \textbf{y}) = \binom{m}{y_1 ... y_k}p_1^{y_1} \dots p_k^{y_k} \prod_{l=1}^{k}\prod_{j=1}^{k}\nu_{l_j}^{y_l y_j}$$ where in particular $m=\sum_{i_1}^{k} y_i$, while $p_1...p_k$ is the vector of probabilities associated to each one of the k categories and $\nu_{l_j}$ for $j,l \in {1,...,k}$ and $j<l$.
Can somebody help me? Also just defining what moment definition means. thanks in advance!