Panel (a): The level of X-ray weakness, Δα_ox, compared to the expected value from Equation (3) of Pu et al. (2020). As in, e.g., Gibson et al. (2009) we find that the BAL quasars (open purple) are generally significantly X-ray weaker than the typical (filled cyan) and mini-BAL (filled orange) quasars, probably due to heavy absorption by the gas associated with the BAL outflow. Panel (b): The distribution of log(count-flux ratio) for the BAL, mini-BAL, and typical quasars. The arrows depict count-flux ratios in which one epoch is not X-ray detected, where the direction of the arrow represents whether the non-detection is in the numerator (left) or denominator (right). A Shapiro–Wilk test indicates that the BAL and mini-BAL quasar distributions are consistent with a Gaussian distribution, and we find that that the intrinsic dispersions of the X-ray variability amplitudes for the three populations are statistically consistent. Panel (c): Count-flux ratio as a function of timescale, Δt, for BAL quasars and the matched sample of typical quasars from T20 along with their respective median 1σ error bars. Both quasar samples have been down-sampled using the method outlined in Section 4.1 of T20. Panel (d): The same as panel (c), however we depict the mini-BAL quasars.