Regression analysis of the final selected sample (Γ = 1.7−2.8, k δ_obs = 0.6). The division of the sample in logarithmic redshift bins of size Δlog(z)=0.06 yields 17 bins with more than five objects. The results for two fitting methods adopted, emcee and linmix, are plotted in (dashed) lime green and magenta, respectively. For each sub-plot, the inset shows the distribution of off-axis angles Θ for the sources in the redshift bin. Sources with at least one observation with SN < 5 in the soft band are marked with a red point. For each redshift bin, we list the median redshift ⟨z⟩, the number of data points N and dispersion and slope from the emcee algorithm.