= Solution
The reduced model `fit3` retains malignancy but omits both site indicators. The <analysis of deviance for nested generalized linear models> therefore tests
$$
H_0:\beta_B=\beta_C=0
$$
against the additive model with a site effect. The likelihood-ratio statistic is the reduction in deviance,
$$
7.4923-0.8505=6.6418.
$$
The full model adds two parameters, so under $H_0$ this is approximately $\chi^2_2$. Since
$$
6.6418>\chi^2_{2,0.95}=5.9915
$$
but is less than the $0.99$ quantile $9.2103$, the p-value is between $0.01$ and $0.05$; in fact, because $\chi^2_2$ has survival function $e^{-x/2}$,
$$
p=e^{-6.6418/2}\approx0.0361.
$$
We reject the no-site-effect null at the $5\%$ level and find evidence that survival differs by site after adjustment for malignancy.
Back to article page
