Large-scale bias as a function of redshift (panels) and obscuration (markers in accordance with the legend). The vertical error bars correspond to the 16% and 84% quantiles of the best-fit bias using bootstrap resampling. The horizontal error bars show the 1σ of the N_H distribution of the sample. Measurements from X-ray AGN clustering in terms of the HR at comparable redshifts are shown as gray markers in accordance with the legend (Gilli et al. 2009; Allevato et al. 2011; Mendez et al. 2016; Koutoulidis et al. 2018). Note that for plotting purposes we assume N_H = 10²¹ cm⁻² (N_H = 10²³ cm⁻²) for the soft (hard) samples and include a slight offset in N_H for visual clarity. With open markers we show the bias based on the correlation length measurement, r₀, assuming a power-law index γ = 1.8, as reported by Gilli et al. (2009), Allevato et al. (2011), Mendez et al. (2016), and Koutoulidis et al. (2018).