Another potentially problematic issue with model averaging is the selection of candidate probability models. As noted by Burnham and Anderson (2002), the construction of the candidate model set involves an element of subjectivity and that “one must recognize a certain balance between keeping the set small and focused on plausible hypotheses, while making it big enough to guard against omitting a very good a priori model”. Furthermore, and of relevance to ecotoxicology, they note that these problems “are particularly serious when little theory is available to guide the analysis”.

5 Responses

  • RESOLUTION

    Respond by saying it appears to be a comment without recommendation but we have hopefully clarified by replacing “Furthermore, and of relevance to ecotoxicology, they note that these problems “are particularly serious when little theory is available to guide the analysis” with:
    
    Similarly, Wheeler and Bailer (2009) point out that the efficacy of model averaging when applied to dose-response modelling is very much dependent on the chosen model space. If the true dose-response model lies within the chosen model space Wheeler and Bailer (2009) conclude that model averaging is superior to other commonly used approaches but may perform poorly otherwise and hence the suggestion that the model space includes a wide variety of model curvatures.

  • I think as Rick and David already indicate, the MS already addresses this somewhat, and yes I agree with David that their comment is inconsistent. It is precisely the fact that there is little theory to guide the choice of SSD distribution that the model averaging approach is useful in the SSD context. Of course construction of the candidate set is subjective, but the selection of a single model is even more so. Providing we are including all readily available distributions that encompass all reasonable alternative “shapes” (but see #31) the model averaging approach vastly improves issues of subjectivity in SSD fitting.

  • This looks to be a comment without a recommendation.Can we simply say that the comment is noted and that the key aspects of the comment are already acknowledged in the ms?

Comments are closed.