Values of the reduced chi-squared (χ_red²) and the Bayes’ factor relative to NLA for TATT, TATT-M (A₂ = 0), TATT-M and TATT-M′. Our TATT-M model, based on an empirical fit to the halo data, is very strongly preferred over NLA on the central galaxy sample. TATT, TATT-M (A₂ = 0) and TATT-M′ (based on empirical fits to the galaxy sample), are also preferred over NLA. There is a strong preference for the TATT-M model.