Use the full set of cancer counts in Table 8.1 (both BGA and CTR locations) to fit a Poisson regression model to the cancer rate data with median age as the covariate. Note that this model ignores the location variable—that is, it treats the effect of (median) age on the cancer rate the same way, regardless of whether the counts are from BGA or CTR. Do you think this makes sense in light of the estimated coefficients you found in Questions 15 and 17?