Request for help!

[Ailton Matheus]

Hello everyone,

I'm Ailton Matheus. I'm new to using BioGeoBEARS and I am using six regions for my analyses: Neotropical (A), Nearctic (B), Palearctic (C), Indomalayan (D), Australasian (E), and Afrotropical (F).

Currently, I am working with Fungi, and my topology has 39 tips. My results suggest that BAYAREALIKE+J is the ideal model for my data. However, looking at the table named "restable_AIC_rellike_formatted.txt", my AIC_wt is 1.00. This seems strange to me and doesn't make sense.

Could there be an error with my data?

I apologize if this question is too obvious, but I'm still learning about this field.

LnL numparams d e j AIC AIC_wt
DEC -84.86 2 0.0047 1.0e-12 0 173.7 1.1e-06
DEC+J -84.16 3 0.0044 1.0e-12 0.0075 174.3 8.1e-07
DIVALIKE -89.39 2 0.0055 1.0e-12 0 182.8 1.2e-08
DIVALIKE+J -88.73 3 0.0050 1.1e-09 0.0084 183.5 8.4e-09
BAYAREALIKE -77.89 2 0.0033 0.011 0 159.8 0.0012
BAYAREALIKE+J -70.14 3 0.0022 1.0e-07 0.028 146.3 1.00

[Ivan Magalhães]

Hi Ailton, this looks correct. AIC weights compare the relative fits of models and always sum up to 1. For instance, if you have two models with equal fit, each would have an AIC weight of 0.5. 

In your case, the AIC weight of the preferred model is not actually 1, but 1 - ( 1.1e-06 + 8.1e-07  ... + 0.0012 ) (the sum of all the alternative models). However, the other models have such small AIC weights that the AIC weight of the preferred model is being rounded to 1.

For more information on AIC weights, read Wagenmakers & Farrell 2004, AIC model selection using Akaike weights.

Best,

Ivan

--
You received this message because you are subscribed to the Google Groups "BioGeoBEARS" group.
To unsubscribe from this group and stop receiving emails from it, send an email to biogeobears...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/biogeobears/a6d85035-06db-4681-b42c-b73b1c5ff4cfn%40googlegroups.com.