Comparing AIC/model averaging in closed population models

[mondulkiri]

I am analysing camera-trap data on leopard to estimate N using closed population models The data set show strong heterogeneity in capture probabilities that is mostly, but not entirely, explained by gender.

I have ran models with gender as a grouping attribute (e.g. as described in ch 11.7 in MARK gentle guide) and also used Pledger mixture models (with two groups) to estimate N.

Is it valid to compare AICs between these two model types and get a model averaged estimate of N?

Thanks

Tom

[cooch]

I am analysing camera-trap data on leopard to estimate N using closed population models The data set show strong heterogeneity in capture probabilities that is mostly, but not entirely, explained by gender.

I have ran models with gender as a grouping attribute (e.g. as described in ch 11.7 in MARK gentle guide) and also used Pledger mixture models (with two groups) to estimate N.

Is it valid to compare AICs between these two model types and get a model averaged estimate of N?

Thanks

Tom

As noted in Chapter 14, it is legitimate to model average across data types that are based on the same likelihood, but not across those based on different likelihoods. So, for example, you could not average between models based on the full likelihood and those based on the Huggins-Ahlo likelihood. Hence, AICc cannot be compared between these models, for the same reason. However, I believe standard closed models (with N in the likelihood) and standard closed models with heterogeneity (Pledger mixture models), again with N in the likelihood, can be averaged.