Notes. The first column lists the probes and models under consideration in this work. The second column list the change in the χ² value at the maximum of the posterior compared to the fiducial results of Asgari et al. (2021) and Heymans et al. (2021). The quoted uncertainty is the scatter between optimisation runs. Columns 3−5 list the three model selection criteria considered in this work: DIC (Eq. (1)), WAIC (Eq. (2)), and change in the evidence (Eq. (3)), with the uncertainties due to the stochasticity of nested sampling estimates. The last three columns list the model probabilities based on the three selection criteria with respect to the fiducial, flat ΛCDM model, as defined in Sect. 2.3. The relative uncertainty on the model probabilities are of the order of 10−20% but for clarity we do not quote them here.