multinomial logistic regression-warning of zero frequencies?

Moderators: statman, Analyst Techy, andris, Fierce, GerineL, Smash

Posts: 4
Joined: Sun Jun 29, 2008 9:23 pm

multinomial logistic regression-warning of zero frequencies?

Postby zannaclay » Sun Jun 29, 2008 9:40 pm

I am trying to do a multinomial logistic regression to see if the type of food ( high,med, low prefernce) eaten by the primate that i am studying can be predicted by the proportion of different call types they give. there are 5 call types, some call types only given when they eat something they really like and some only fgiven when they dont like the food

when i try and run it, i get warning signs saying there are many cells with zero frequencies ( ie- dependent v levels by subpopulations)

why is this, and how to recify? for many of my cases,there are zeros for some call types ( ie- they do not produce call 5 when they like a food).

any advice greatly appreciated!
Posts: 1
Joined: Thu May 02, 2013 8:45 pm

Re: multinomial logistic regression-warning of zero frequenc

Postby agnese » Thu May 02, 2013 9:01 pm

I am having the same problem, but it is very stange because I have not inserted any factors, only 3 (metric) covariates. I also tried running the analysis with some factors but checked all the crosstabs and had no 0 expected frequencies. I have also tried running an example from Tabachnick anf Fidell (2001) Using Multivariate Statistics. I used their dataset and copied their syntax and the same message appears. Have you found a solution or have you just ignored the warning? Thanks!!

Who is online

Users browsing this forum: No registered users and 1 guest