Cell counts {Y i } are independent Poisson random variables, with i = E(Y i ).

Cell counts {Y_i} are independent Poisson random variables, with µ_i = E(Y_i). Consider the Poisson loglinear model

log µ = X_a θ_a, where µ = (µ₁,..., µ_N).

Using arguments similar to those in Section 14.2, show that the large-sample covariance matrix of θ̂_a can be estimated by [X’_a diag(µ̂)X_a]^–1, where µ̂ is the ML estimator of µ.