In a detailed model comparison, we assessed the model results on consumption, agricultural production, commodity prices, land-use change and greenhouse gas emissions, and identified feedbacks in the global agricultural system.
Data from the FACE studies, including this one, are being used in a multi-model comparison to assess land surface model structure and variation in predictive capacity, which will result in future model improvement (e.g. De Kauwe et al. 2013; Walker et al. 2014).
To address this problem, Spiegelhalter et al.[ 30] suggest DIC as a generalization of the Akaike information criterion (AIC) which can be used as a Bayesian approach for model comparison and to assess the adequacy of hierarchical models.
These comparisons assessed the improvement in model fit of including the third measure.
The other level of inference is the model comparison so as to assess the more plausible model.
We assessed additive model comparison, heterozygote comparison and dominant model comparison by tumor type, ethnicity, publication year, control source, sample size and HWE in controls.
A formal model comparison would be possible by assessing model predictive ability using cross-validation, but this was beyond the aim of our study.
Since we knew the ground truth (i.e., which model had generated the data) this allowed us to assess the veracity of model comparison.
Hence, in the context of model comparison, scoring rules provide a diagnostic approach to assessing the predictive performance of a model.
These conditions were then modelled using a finite difference mathematical model and the accuracy of the model assessed by comparison with the experimental results.
