Solution
= Solution
Time is exposure: doubling the public duration should double the expected count without changing the viewing rate. Modeling the rate is therefore preferable to treating time as an additive linear predictor. In a <Poisson regression> the exact implementation should be
$$
\log\mathbb E(N_i)=\log t_i+\beta_0+\beta_1I_i,
$$
using $\log t_i$ as an offset; directly supplying noninteger $N_i/t_i$ without exposure weights is not generally likelihood-equivalent.