tag_model_avg: Model Averaging for Instantaneous Rates Tag Return Models
Description
Calculates model averaged estimates of instantaneous fishing, natural and total mortality, and survival rates for
instantaneous rates tag return models (Hoenig et al. (1998) and Jiang et al. (2007)).
Usage
tag_model_avg(..., global = NULL)
Arguments
…
model object names separated by commas
global
specify global model name in quotes. If the global model is the first model included
in the list of candidate models, this argument can be ignored.
Value
List containing model summary statistics, model-averaged estimates of fishing, natural, tag, and total mortality, and
survival and their weighted and uncondtional standard errors .
Details
Model estimates are generated from functions irm_cr and irm_h.
Averaging of model estimates follows the procedures in Burnham and Anderson (2002).
Variances of parameters are adjusted for overdispersion using the c-hat estimate from the global model
: sqrt(var*c-hat). If c-hat of the global model is <1, then c-hat is set to 1. The c-hat is used to calculate the quasi-likelihood AIC and AICc
metrics for each model (see page 69 in Burnham and Anderson(2002)). QAICc differences among models are calculated by
subtracting the QAICc of each model from the model with the smallest QAICc value. These differences are used to calculate
the Akaike weights for each model following the formula on page 75 of Burnham and Anderson (2002). The Akaike weights are
used to calculate the weighted average and standard error of parameter estimates by summing the product of the model-specific Akaike weight and parameter estimate
across all models. An unconditional standard error is also calculated by
sqrt(sum(QAICc wgt of model i
* (var of est of model i
+ (est of model i - avg of all est)^2))).
References
Burnham, K. P. and D. R. Anderson. 2002. Model selection and multimodel inference : A Practical Information-Theorectic Approach, 2nd edition. Spriner-Verlag, New York, NY. 488 p.
# NOT RUN {## This is a typical specification, not a working example# }# NOT RUN {tag_model_avg(model1,model2,model3,model4,model5,model6,model7,global="model7")
# }