GLM - Scale Deviance

Hi every one,



In the text on GLM by Anderson et al., the authors state that one way to estimate the scale parameter is by using the total deviance estimator:

Scale parameter=D/(n-p) (p.19)



At page 38, the deviance test based on Chi-Square distribution uses the scale deviance, D/scale parameter.

Combining those two formulas, we get that the scale deviance is equal to (n-p).



I'm I right? I find this a little bit strange that scale deviance is equal to the degree of freedom...



Thanks in advance for your help,



Emi

GLM - Scale Deviance