The numbers of mutations were described by a Poisson regression model with parameter
estimates obtained by the method of maximum likelihood [37]. The statistical model
included the effects of age, time after irradiation, cell type, all two-factor interactions, and
the three-factor interactions. Statistical tests of differences used the likelihood ratio test.
Comparisons among ages for mice not irradiated (mock) were regarded as a priori statistical
tests; other comparisons among groups were Bonferroni adjusted. Computations were made
using SAS PROC GENMOD (Version 9.2, SAS Institute, Cary, NC).