For rate data, the Poisson GLM with identity link is

a. Since the model has form at +tx, argue that it is equivalent to a Poisson GLM for the response totals at the various levels of x, using identity link with t and tx as explanatory variables and no intercept or offset terms.

b. Fit this model to the grouped data in Table 4.3, using average width scores. Compare results, including interpretations, goodness of fit, and residual analyses, to those obtained with the log link.

= a + Bx. t