1
votes

I was comparing two models using the AIC. However, I realized that both AIC values are too small (-4752.66, and the other is close to that). I was wondering if that is normal or I did something wrong while calculating it.

1

1 Answers

1
votes

Its ok to have negative aic values! (Ref-https://stats.stackexchange.com/questions/84076/negative-values-for-aic-in-general-mixed-model)

As the values are too close to each other, the one with smaller delta is your choice. If both deltas are smaller than two, look for evidence ratio . Evidence ratio smaller than 2.7 indicates that booth models are good ->(serch for "comparing models" at : http://theses.ulaval.ca/archimede/fichiers/21842/apa.html#d0e5831).

IF this is your case, you can use model averaging->(Symonds, Matthew RE, and Adnan Moussalli. "A brief guide to model selection, multimodel inference and model averaging in behavioural ecology using Akaike’s information criterion." Behavioral Ecology and Sociobiology 65.1 (2011): 13-21.)