Page 1 of 1

cumulative logit link missing?

PostPosted: Wed Apr 22, 2015 2:48 pm
by ehileman
Has the cumulative logit link been removed from the latest MARK build (version 8.0, Mon, 06 Apr 2015 20:34:38 GMT)? I was looking to use it for a Pradel survival and lambda analysis and it is no longer listed in the drop down menu when using Parm-specific link functions. Am I missing something? Thanks in advance for any insight into this matter.

Cheers,

Eric

Re: cumulative logit link missing?

PostPosted: Wed Apr 22, 2015 2:53 pm
by cooch
ehileman wrote:Has the cumulative logit link been removed from the latest MARK build (version 8.0, Mon, 06 Apr 2015 20:34:38 GMT)? I was looking to use it for a Pradel survival and lambda analysis and it is no longer listed in the drop down menu when using Parm-specific link functions. Am I missing something? Thanks in advance for any insight into this matter.

Cheers,

Eric


Never was in the drop-down list. You have to physically write in CLogit(xx) yourself. See - sidebar- starting on p. 49 of Chapter 6 of the MARK book.

Re: cumulative logit link missing?

PostPosted: Wed Apr 22, 2015 3:00 pm
by ehileman
Thanks for the quick reply, Evan! That makes more sense--I guess it has been longer than I thought since the last time I've needed to use it.

Cheers,

Eric

Re: cumulative logit link missing?

PostPosted: Wed Apr 22, 2015 3:10 pm
by cooch
ehileman wrote:Thanks for the quick reply, Evan! That makes more sense--I guess it has been longer than I thought since the last time I've needed to use it.

Cheers,

Eric


Sure. CLogit can be handy (and I suspect you're interested in using it to enforce the logical constraint between lambda and S). Just remember to use the Clogit trick 'sparingly' -- see sidebar on p. 9 of the Pradel model chapter. Also, as noted in the sidebar, for anything not included in the CLogit constraint, you'll want to make sure you set the appropriate link (log or otherwise).

Re: cumulative logit link missing?

PostPosted: Wed Apr 22, 2015 3:19 pm
by ehileman
Sure. CLogit can be handy (and I suspect you're interested in using it to enforce the logical constraint between lambda and S). Just remember to use the Clogit trick 'sparingly' -- see sidebar on p. 9 of the Pradel model chapter. Also, as noted in the sidebar, for anything not included in the CLogit constraint, you'll want to make sure you set the appropriate link (log or otherwise).

Your suspicion is correct. Thanks for the reminder and additional information. Cheers! -Eric