C27. Quasilikelihood

Topic: Underdispersion, positive responses, and quasi-likelihood · more questions on this topic
‹ C23 C28 ›

What is a quasilikelihood model? What must be specified instead of a full response distribution?

Worked answer (click to show)

Definitions/notation. A quasilikelihood model specifies a mean \mu_i=g^{-1}(x_i^\top\beta) and variance \phi V(\mu_i), rather than a full response distribution. For independent responses its quasi-score is built from (Y_i-\mu_i)\{\phi V(\mu_i)\}^{-1}\partial\mu_i/\partial\beta, whose expectation is zero when the mean model is correct.