A binomial GLM i = ( j j x ij ) with arbitrary inverse link

A binomial GLM π_i = Φ(∑_j β_j x_ij) with arbitrary inverse link function Φ assumes that n_iY_i has a bin(n_i, π_i) distribution. Find w_i in (4.27) and hence cov͡ (β̂). For logistic regression, show that w_i = n_i π_i(1 – π_i).