For the data given in Table 4.5 of Agresti (and provided also in file heart.csv), fit the model: log(µij/tij ) = a + ß1ai + ß2?j , i, j = 1, 2 where: Yij : Denote the number of deaths for age ai and valve type ?j , with expected value µij for total time at risk tij (months). a is the indicator variable for age, with a1 = 0 for the younger group, and a2 = 1 for the older group. ? is the indicator variable for valve type, with ?1 = 0 for aortic; and ?2 = 1 for mitral. (a) Find the 95% profile likelihood Confidence Intervals for ß1 and ß2. (b) Compare the goodness of fit of the model when the canonical link is used and when the identity link is used.