is also 2 for the multi-dimensional versions and the expected proportion of successes is the same for all components. Their multivariate version is therefore inflexible and not useful for modeling data like those presented in the example in Section 5