[R] [FORGED] Re: SE for all levels (including reference) of a factor atfer a GLM

[Marc Girondot]

Le 16/02/2018 à 10:24, Peter Dalgaard a écrit :
> To give a short answer to the original question:
>
>> On 16 Feb 2018, at 05:02 , Rolf Turner <r.turner at auckland.ac.nz> wrote:
>>
>> In order to ascribe unique values to the parameters, one must apply a "constraint".  With the "treatment contrasts" the constraint is that
>> beta_1 = 0.
> ...and consequently, being a constant, has an s.e. of 0.
>
>
It makes sense !

Thanks for the precision !

Marc