reference for the multinomial logit link

[markmiller]

I am attempting to locate a reference for the multinomial logit link. So far the only reference I have found mentioned is:

Almeras, T. 1996. Models de capture-recapture multisites: integration de models contraints dans un meme cadre souple et convivial, realisation d'un logiciel de traitement de donnees sous Matlab. DEA report, Montpellier, Montpellier II University.

Is this the standard reference? If not, could someone provide the citation for the standard reference? Either way, please consider emailing a pdf of the reference to me if it is not widely accessible.

I have seen a request not to cite the Program MARK 'Gentle Introduction'.

Thank you,

Mark

mark.wayne.miller@gmail.com

[cooch]

I am attempting to locate a reference for the multinomial logit link. So far the only reference I have found mentioned is:

Almeras, T. 1996. Models de capture-recapture multisites: integration de models contraints dans un meme cadre souple et convivial, realisation d'un logiciel de traitement de donnees sous Matlab. DEA report, Montpellier, Montpellier II University.

Is this the standard reference? If not, could someone provide the citation for the standard reference? Either way, please consider emailing a pdf of the reference to me if it is not widely accessible.

I have seen a request not to cite the Program MARK 'Gentle Introduction'.

Thank you,

Mark

mark.wayne.miller@gmail.com

You don't need a citation for a link transformation -- any more than you would look for a citation for (say) a square-root of something, or a log of something, or anything else. If you use a logit link, would you cite something? Probably not. Same applies for a multinomial link. The multinomial logit link has been around as long as there have been people working with multinomial response models, which is a while.

So, don't sweat it. Simply say you used a multinomial link, and move on.