Observational data sets generated by complex processes are common in ecology. Traditionally these have been very challenging to analyze because of the limitations of available statistical tools. This seems to be changing, and these are exciting times to be involved with ecological statistics, not just because of the neo-Bayesian revival but also because of the proliferation of computationally intensive methods in general. It is now possible to fit much richer models to observational data than in the relatively recent past, which in turn has stimulated much interest in how to evaluate and compare such models. In such an immature, vibrant, and rapidly growing field, not everyone is going to agree on the best way to do things. This is reflected in the contrast of opinions offered by the discussants. Each offers a thoughtful and thought-provoking critique of our work that reflects the current thinking in a non-negligible segment of the ecological data analysis community. We want to thank them for their insights.