We compare parameter point and interval estimates based on the symmetric bounded loss function, as used in the Add-my-Pet collection on animal energetics, with the maximum likelihood method for number of surviving individuals as function of time. The aging module of Dynamic Energy Budget theory is used to generate Monte Carlo data sets. The simulations show that estimates based on the symmetric loss function give almost the same results in terms of point as well as interval estimates, compared to maximum likelihood estimation, while this loss function avoids the need to model the stochastic component of data sets. For most data types on energetics, we don't have such stochastic models, so maximum likelihood methods cannot be used. Our findings support the view that model plasticity dominates interval estimates, rather than the detailed structure of the stochastic component.
- Dynamic energy budget theory
- Maximum likelihood
- Profile-based interval estimates
- Survival data
- Symmetric bounded loss function