Given your answer to Question 32, fit a different Poisson log- linear model to the CTR data that you believe may more accurately reflect the relationship between cancer incidence rates and age. The models log(λ) = log(t) + β0 + β1x2 and log(λ) = log(t) + β0 + β1x + β2x2 are both still generalized linear models. The functions are linear in terms of the model coefficents, yet allow a nonlinear relationship between x and log(λ). What does the deviance statistic suggest about the fit of your new model?