Posterior predictive of the Bayesian analysis using KiDS-1000 tomographic shear correlations and a relative power, f_δ(k, z), averaged inside the three redshift bins Z₁ = [0, 0.3], Z₂ = [0.3, 0.6], and Z₃ = [0.6, 2]. For each tomographic bin combination (ij), the panels with labels ‘z–ij’ show the posterior model constraints as light blue (95% CI) regions and dark blue (68% CI) regions about the median for either θ ξ₋^(ij)(θ) (lower left triangle) or θ ξ₊^(ij)(θ) (upper right triangle), both in units of ${10}^{-4}\mathrm{arcmin}$ and as function of lag θ. Black points with error bars (1σ) are the KiDS-1000 data points. The red lines correspond to the ΛCDM reference power spectrum with S₈ ≈ 0.73, the solid green lines ‘SIM-THS17’ correspond to the prediction by Takahashi et al. (2017), see Sect. 5.5, for S₈ ≈ 0.79. Errors of ξ_±^(ij)(θ) are correlated between θ-bins and tomographic bins, marginal errors due to lensing kernel and IA uncertainties are not included here (adding another ∼10% to CIs). Conflicts with the data are visible for θ ξ₊^(ij) in z–22 and to a lesser degree for z–12 to z–15. Figure A.4 shows a random realisation of the reference model.