Re: likelihood model for missing SEX data

From: Nick Holford n.holford@auckland.ac.nz Subject: Re: [NMusers] likelihood model for missing SEX data Date: Thu, 06 Jul 2006 00:33:35 +1200 Jakob, I agree with your comments (more or less). The simple mixture model approach does however work quite well (N=1 example) provided the mixing fraction is FIXED to the a priori best guess obtained from the non-missing SEX ratio. More complex predictions of SEX e.g. based on WT are of course very sensible but for the purposes of method illustration the model I suggested was kept simple. I will shortly post a further comment to the list with an illustration of how to use the joint likelihood method that Anthe asked about. This can be combined with the mixture model method as well. I have not yet figured out how to do this with the individual mixing probability approach that you mention. You can show us how to do this when I have posted the codes (and data) for the other methods :-) Nick